An SF6-N2 partial discharge decomposition product detection device and a GIS partial discharge evaluation method

Abstract

The present invention relates to the technical field of insulation gas state detection of electrical equipment, specifically a detection device for SF6-N2 partial discharge decomposition products and a detection and GIS partial discharge evaluation method, and particularly relates to a technology for detecting SF6-N2 decomposition components by a novel Ag2O-InN gas sensor. The present invention is based on the GRNN neural network method of MATLAB, sets N characteristic gases at M concentration levels, and the response value data points are M N ×6. Through the resistance change of the array-type Ag2O-InN resistive gas sensor, the types and contents of SF6-N2 decomposition products are detected, and then the occurrence and severity of partial discharge are judged. The present invention is based on the detection of GIS partial discharge gas and the evaluation of partial discharge severity by the GRNN neural network of MATLAB, which can realize the monitoring of GIS partial discharge faults and ensure the safe and reliable operation of GIS.

Description

Technical Field

[0001] The present invention relates to the technical field of electrical equipment insulating gas state detection, in particular to a SF6-N2 partial discharge decomposition product detection device and a detection and GIS partial discharge evaluation method. Background Art

[0002] Sulfur hexafluoride (SF6) gas is a gas with very stable chemical properties. As an excellent insulating and arc-extinguishing medium, it is widely used in various electrical equipment. However, SF6 gas is a greenhouse gas listed in the Kyoto Protocol and the Bali Roadmap. Its greenhouse effect is 23,900 times that of the same amount of CO2 gas, and SF6 gas can exist stably in the atmosphere for up to 3,200 years. In order to solve the increasingly prominent greenhouse effect and low-temperature liquefaction problems of SF6 gas, mixed insulating gases such as SF6 / N2 and SF6 / CF4 are considered to be the most promising alternative media. Due to the harmlessness and low price of N2, using SF6 / N2 mixed gas instead of high-purity SF6 gas as an insulating medium is a trend in the development of green electricity.

[0003] SF6-N2 gas mixture can reduce the use of SF6 while ensuring insulation performance, and is an excellent alternative insulating gas. In the case of partial discharge (PD) in gas insulated switchgear (GIS), the characteristic quantities used to judge the entire PD process, early PD process and severe PD process are NO2, SO2F2 and (SOF2+SOF4) respectively. It is necessary to explore a detection method for these decomposition components and GIS partial discharge evaluation method to judge the occurrence and severity of PD. Summary of the invention

[0004] In view of the deficiencies in the prior art, the object of the present invention is to provide a device and method for detecting SF6-N2 partial discharge decomposition products. The present invention realizes online detection of the types and concentrations of partial discharge decomposition products of SF6-N2 in gas insulated switchgear, and then determines the occurrence and severity of partial discharge.

[0005] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0006] A GIS partial discharge evaluation method for SF6-N2 partial discharge decomposition products comprises the following steps:

[0007] The gas to be tested of GIS is introduced into the SF6-N2 partial discharge decomposition product detection device, and the gas to be tested is detected by the array-type Ag2O-InN resistive gas sensor to obtain the response data of the resistance change;

[0008] Construct a response value matrix based on the response data of the resistance change;

[0009] Construct a neural network, input the response value matrix into the trained neural network, and obtain the prediction result of the partial discharge severity level; wherein, the partial discharge severity level is three levels of 0, 1, and 2, corresponding to no partial discharge, early partial discharge, and severe partial discharge respectively.

[0010] Preferably, the neural network is a GRNN neural network based on MATLAB; the training of the GRNN neural network based on MATLAB includes the following steps:

[0011] Construct a training set, the training set includes the response value matrix of the Ag2O-InN resistive gas sensor and the matrix of the corresponding partial discharge severity level; wherein, when the concentration values of NO2, SOF2, SO2F2, and SOF4 are greater than zero, partial discharge occurs; when the concentration of SO2F2 increases rapidly with time, the partial discharge is in the early stage; when the output concentrations of SOF2 and SOF4 are much greater than those of NO2 and SO2F2, severe partial discharge occurs;

[0012] Use the sensor response value matrix of the training set as the input of the GRNN neural network, and the matrix of the partial discharge severity level as the output of the GRNN neural network to train the neural network;

[0013] Randomly select a part of the training set as the test set, calculate the absolute error and relative error between the predicted value and the actual value, and verify the performance of the neural network.

[0014] Preferably, the construction of the training set includes:

[0015] Set 8 concentration levels of 2, 4, 6, 8, 10, 14, 16, and 20 μL / L for the four characteristic gases of NO2, SO2F2, SOF2, and SOF4, then the mixing ratio of the mixed gas is 8 4 species, and the number of response value data points is 8 4 ×n, where n is the number of array-type Ag2O-InN resistive gas sensors.

[0016] Preferably, the GRNN neural network includes an input layer, a hidden layer, and an output layer:

[0017] The number of neurons in the input layer is N i , the number of neurons in the hidden layer is N h , and the number of neurons in the output layer is N o ;

[0018] The determination method of the number of neurons in the hidden layer is: N o <Nh <N i , or N h =2N o / 3 + 2N i / 3, or N h =2N o 。

[0019] Preferably, the number of neurons in the hidden layer is: N h =N s / α * (N i +N o ); where N s is the number of samples in the training set, and α = 2 - 10.

[0020] A detection device for a GIS partial discharge evaluation method using SF6-N2 partial discharge decomposition products, comprising:

[0021] A GIS circuit breaker gas chamber, which is a gas chamber for the SF6-N2 mixed gas to be measured;

[0022] A sampling chamber, which is connected to the GIS circuit breaker gas chamber respectively through an intake pipe and an intake valve, and an exhaust pipe and an exhaust valve;

[0023] Six Ag2O-InN resistive gas sensors, combined into an array, are installed in the sampling chamber to obtain a response value matrix according to the resistance change;

[0024] An electrochemical workstation, which acquires the response value matrix and analyzes to obtain the types and contents of SF6-N2 decomposition products.

[0025] Preferably, it further comprises:

[0026] A data analysis unit, which obtains a prediction result of the partial discharge severity level through a GRNN neural network based on MATLAB according to the response value matrix;

[0027] A data storage unit, which is used to store the prediction result of the data analysis unit.

[0028] Preferably, the Ag2O-InN resistive gas sensor comprises six Ag2O-InN sensors, and the six Ag2O-InN sensors are respectively made of Ag2O-InN with a silver content of 5%, and the gas-sensitive films of the six Ag2O-InN sensors are 4, 6, 8, 10, 12, and 14 layers respectively, and the thickness of each layer of the gas-sensitive film is 300 nm.

[0029] Preferably, it further comprises a constant temperature control device; the temperature of the constant temperature control device is determined by the heating voltage V hControl: When the ambient temperature is higher than 20°C, the preheating time shall be not less than 25 minutes; when the ambient temperature is lower than 20°C, the preheating time shall be not less than 35 minutes.

[0030] Preferably, it further includes a cleaning pipeline and a cleaning pipeline pump; the cleaning pipeline pump is turned on after the gas detection is completed, and the cleaning gas enters the sampling chamber through the auxiliary valve, forcing the residual gas to be discharged.

[0031] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0032] The present invention can on-line detect the types and contents of PD decomposition products in GIS, and conveniently judge PD faults. The present invention realizes on-line detection of the types and concentrations of partial discharge decomposition products of SF6-N2 in gas-insulated switchgear, and then realizes the judgment of the occurrence and severity of partial discharge; the present invention evaluates the degree of GIS partial discharge based on the GRNN neural network of MATLAB, realizes the accuracy of GIS partial discharge detection, and ensures the reliability of GIS operation. Description of the Drawings

[0033] Figure 1 It is a schematic structural diagram of a partial discharge decomposition product detection device of the present invention;

[0034] Figure 2 It is a structural diagram of the GRNN neural network of the present invention.

[0035] Description of the reference numerals: 1 - GIS circuit breaker chamber, 2 - intake pipe, 3 - sampling chamber, 4 - exhaust pipe, 5 - intake valve, 6 - outlet valve, 7 - Ag2O-InN resistive gas sensor, 8 - electrochemical workstation, 9 - data analysis unit, 10 - data storage unit, 11 - auxiliary valve, 12 - cleaning gas, 13 - cleaning pipeline, 14 - constant temperature control device. Detailed Embodiments

[0036] The following describes the detailed embodiments of the present invention in detail, but it should be understood that the protection scope of the present invention is not limited by the detailed embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention. The experimental methods in the embodiments of the present invention are all conventional methods unless otherwise specified.

[0037] Embodiment 1

[0038] The present invention provides a device for detecting partial discharge decomposition products of SF6-N2 and a method for detecting and evaluating GIS partial discharge. The schematic diagram of the device for detecting partial discharge decomposition products of SF6-N2 is as Figure 1 shown, and the structural diagram of the GRNN neural network is as Figure 2 shown.

[0039] S1: Introduce the gas to be measured in the GIS into the SF6-N2 partial discharge decomposition product detection device, and detect the gas to be measured through the array-type Ag2O-InN resistive gas sensor to obtain the response data of the resistance change.

[0040] S2: Construct a response value matrix based on the response data of the resistance change.

[0041] S3: Construct a GRNN neural network based on MATLAB, input the response value matrix into the trained GRNN neural network, and obtain the prediction result of the partial discharge severity level; among them, the partial discharge severity level is divided into three levels: 0, 1, and 2, corresponding to no partial discharge, early partial discharge, and severe partial discharge respectively.

[0042] Specifically:

[0043] Introduce a mixed gas containing a certain concentration of NO2, SO2F2, SOF2, and SOF4 into the sampling chamber. There are 8 concentration levels of 2, 4, 6, 8, 10, 14, 16, and 20 μL / L for the four characteristic gases. Then the ratio of the mixed gas is 8 4 types, and the response value data points are 8 4 ×6.

[0044] Establish a GRNN neural network, and use 70% of the measured data in the data storage unit as the sample set and 30% of the data as the test set to train the network. The model has 6 inputs and 3 outputs. The response value X of the i-th sensor i =[r1,r2,r3,……,r n T , i = 1, 2, 3, ……, 6; the response value matrix of the measured sensor is X = [X1, X2, X3, ……, X6], and the matrix X is used as the input of the neural network; the matrix Y of the measured partial discharge severity level is Y = [Y1, Y2, Y3], and the matrix Y is used as the output of the neural network.

[0045] Furthermore, the number of neurons in the input layer is 6, the number of neurons in the output layer is 3, and the number of neurons in the pattern layer is the number of training set samples.

[0046] Specifically, the transfer function of the neurons in the pattern layer is:

[0047]

[0048] where X is the sensor response input to the network, X m is the test set sample corresponding to the m-th neuron, and σ is the smoothing factor.

[0049] Specifically, the transfer function of the summation layer is:​

[0050]

[0051] Further, the output of the neural network is as follows:

[0052]

[0053] Further, taking the sensor response value data as the input of the trained neural network, the severity level of partial discharge can be output, and the severity of partial discharge can be predicted.

[0054] Embodiment 2

[0055] A detection device for a GIS partial discharge evaluation method applying SF6-N2 partial discharge decomposition products, comprising: a GIS circuit breaker chamber 1, which is a SF6-N2 mixed gas chamber to be measured; a sampling chamber 3, which is respectively communicated with the GIS circuit breaker chamber 1 through an intake pipe 2 and an intake valve 5, and an exhaust pipe 4 and an exhaust valve 6; 6 Ag2O-InN resistive gas sensors 7, combined into an array and installed in the sampling chamber 3, and a response value matrix is obtained according to the resistance change; an electrochemical workstation 8, which acquires the response value matrix and analyzes to obtain the types and contents of SF6-N2 decomposition products; a data analysis unit 9, which obtains the prediction result of the partial discharge severity level through a GRNN neural network based on MATLAB according to the response value matrix; a data storage unit 10, which is used to store the prediction result of the data analysis unit 9. A constant temperature control device 14, the temperature of the constant temperature control device 14 is controlled by a heating voltage V h When the ambient temperature is greater than 20°C, the preheating time is greater than 25 minutes; when the ambient temperature is lower than 20°C, the preheating time is greater than 35 minutes. A cleaning pipeline 13 and a cleaning pipeline pump, the cleaning pipeline pump is turned on after the gas detection is completed, and the cleaning gas 12 enters the sampling chamber 3 through the auxiliary valve 11, forcing the residual gas to be discharged.

[0056] Further, the Ag2O-InN resistive gas sensor includes 6 Ag2O-InN sensors, and the 6 Ag2O-InN sensors are respectively made of Ag2O-InN with a silver content of 5%, and the gas-sensitive films of the 6 Ag2O-InN sensors are 4, 6, 8, 10, 12, and 14 layers respectively, and the thickness of each gas-sensitive film is 300 nm.

[0057] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention is also intended to include these changes and modifications.

Claims

1. A GIS partial discharge evaluation method for SF6-N2 partial discharge decomposition products, characterized in that It includes the following steps: Introduce the gas to be measured of GIS into the SF6-N2 partial discharge decomposition product detection device, and detect the gas to be measured through the array-type Ag2O-InN resistive gas sensor to obtain the response data of resistance change; Construct a response value matrix according to the response data of resistance change; Construct a neural network, input the response value matrix into the trained neural network, and output the evaluation result of the partial discharge severity level; among them, the partial discharge severity level is three levels of 0, 1, and 2, corresponding to no partial discharge, early partial discharge, and severe partial discharge respectively; The neural network is a GRNN neural network based on MATLAB; the training of the GRNN neural network based on MATLAB includes the following steps: Construct a training set, and the training set includes the response value matrix of the Ag2O-InN resistive gas sensor and the matrix of the corresponding partial discharge severity level; among them, if the concentration values of NO2, SOF2, SO2F2, and SOF4 are greater than zero, partial discharge occurs; if the concentration of SO2F2 increases rapidly with time, the partial discharge is in the early stage; if the output concentrations of SOF2 and SOF4 are greater than NO2 and SO2F2, severe partial discharge occurs; Use the sensor response value matrix of the training set as the input of the GRNN neural network, and the matrix of the partial discharge severity level as the output of the GRNN neural network to train the neural network; Randomly select a part of the training set as the test set, calculate the absolute error and relative error between the predicted value and the actual value, and verify the performance of the neural network; The construction of the training set includes: Four characteristic gases, namely NO2, SO2F2, SOF2, and SOF4, are set at 8 concentration levels of 2, 4, 6, 8, 10, 14, 16, and 20 μL / L. Then, there are 8 4 kinds of mixing gas ratios, and the response value data points are 8 4 ×n, where n is the number of array-type Ag2O-InN resistive gas sensors.

2. The GIS partial discharge evaluation method for SF6-N2 partial discharge decomposition products according to claim 1, characterized in that, The GRNN neural network includes an input layer, a hidden layer, and an output layer: The number of neurons in the input layer is N i , the number of neurons in the hidden layer is N h , the number of neurons in the output layer is N o ; The method for determining the number of neurons in the hidden layer is: N o <N h <N i , or N h =2N o / 3 + 2N i / 3, or N h =2N o .

3. The GIS partial discharge evaluation method for SF6-N2 partial discharge decomposition products according to claim 2, characterized in that, The number of neurons in the hidden layer is: N h = N s / α * (N i + N o ); where N s is the number of training set samples, and α = 2 to 10.

4. A detection device for the GIS partial discharge evaluation method applying the SF6-N2 partial discharge decomposition product described in any one of claims 1-3, characterized in that, It includes: The GIS circuit breaker chamber (1), which is the SF6-N2 mixed gas chamber to be measured; The sampling chamber (3) is connected to the GIS circuit breaker chamber respectively through the inlet pipe (2) and the inlet valve (5), and the exhaust pipe (4) and the outlet valve (6); 6 Ag2O-InN resistive gas sensors (7) are combined into an array and installed in the sampling chamber (3) to obtain a response value matrix according to the resistance change; The electrochemical workstation (8) obtains the response value matrix and analyzes and obtains the types and contents of SF6-N2 decomposition products.

5. The detection device according to claim 4, wherein It also includes: The data analysis unit (9) obtains the evaluation result of the partial discharge severity level through the GRNN neural network based on MATLAB according to the response value matrix; The data storage unit (10) is used to store the evaluation result of the data analysis unit (9).

6. The detection device according to claim 4, characterized in that, The Ag2O-InN resistive gas sensor includes 6 Ag2O-InN sensors, and the 6 Ag2O-InN sensors are respectively made of Ag2O-InN with a silver content of 5%, and the gas-sensitive films of the 6 Ag2O-InN sensors are 4, 6, 8, 10, 12, and 14 layers respectively, and the thickness of each gas-sensitive film is 300 nm.

7. The detection device according to claim 4, wherein It further includes a constant temperature control device (14); the temperature of the constant temperature control device (14) is controlled by a heating voltage V h When the ambient temperature is higher than 20°C, the preheating time is not less than 25 minutes; when the ambient temperature is lower than 20°C, the preheating time is not less than 35 minutes.

8. The detection device according to claim 4, wherein It further includes a cleaning pipeline (13) and a cleaning pipeline pump; the cleaning pipeline pump is turned on after the gas detection, and the cleaning gas (12) enters the sampling chamber (3) through the auxiliary valve (11), forcing the residual gas to be discharged.

Images